Frequency drift on scope

[Conrad Hoffman]

Comparing a frequency reference to some DUT on the scope, you can trigger on the reference and monitor the DUT, or set them up as a Lissajous pattern. Either way, there are other frequencies that show no drift, though the period is obviously different. Simple integer ratios I understand, but what about the frequencies at about 5/6 and 7/6 times the reference? Can anybody explain the math to a math dummy?

[bob91343]

Better to read a text and give it a lot of thought.  It's not rocket science but does requre some pondering to see how the frequencies compare.  That's why the only time I use Lissajous is when I am comparing my counter's reference to my rubidium standard.  If the pattern is stationary, the frequencies are identical.

[IconicPCB]

May be a page from the old transfer oscillator technique.
An audio oscillator driving a step recovery diode /mixer while the scope displays mixing products/ birdies on the screen
find adjacent audio tones and that will give You n and n+1 sub harmonics of the monitored frequency.
From there it is easy to calculate unknown frequency.
This is not quite Lissajous pattern but then no need to visually interpret any relations.

EDIT I suppose you could also look for the stable Lissajous patterns and determine two adjacent tones at which they occur giving n and n+1 sub harmonics.

[Conrad Hoffman]

I've read various texts and just need to go read them again. I'm sometimes slow to absorb this stuff. I understand the formulas for Lissajous patterns but not how to evaluate them for the frequencies that create stable 90 degree circle patterns. I assume similar reasoning applies to a simple drift comparison on the scope. For, say, 1 kHz, I get a stable pattern at about 1166 and 833 Hz, which is where I got the 5/6 and 7/6 numbers.

[IconicPCB]

I hope this helps

https://www.electrical4u.com/lissajous-patterns-of-cro/

[Conrad Hoffman]

I've met the enemy and he is me. Looking at various Lissajous pages, I couldn't convince myself that the waveforms I was seeing were even possible. It turns out you can get anomalous results depending on how the scope triggers and how it does alternate mode, vs chopped, along with delays and time-base setting. Ultimately I set up two signal generators, using the TTL out for triggering on one, to explore possible errors. Also did some Lissajous, all at very low frequencies. And then high. Bottom line is I didn't have as reliable triggering on my reference signal (10 MHz GPSDO) as I needed and sometimes "weird" frequencies would appear to be in sync, when they really couldn't be. Now, back to our regular programming.

[bob91343]

Conrad, what you are experiencing is what we all go through.  It is the experience that gives us the knowledge we need to do this stuff.

How the signals get triggered and the various time bases and the Lissajous patterns and all are part of the electronics learning experience.  The more you do, the more you know.

Our tools are sometimes more sophisticated than they should be for our level.  So you lift yourself up by the bootstraps as it were and learn and learn.  Never give up!

[SilverSolder]

I thought Lissajous patterns were made with the scope in XY mode - i.e. no triggering?

(Note: I have never attempted it)

[Conrad Hoffman]

That's correct. I used them to confirm what I was seeing just looking at drift between the two waveforms. Note that I'm using an ancient boatanchor scope, a Tek 545B, and I'm not even sure the horizontal bandwidth input is high enough to do a 10 MHz Lissajous (didn't try it), but it has a lot of trigger options for normal operation. Yes, I could wheel the 350 MHz Tek 7000 mainframe over nearer to the bench, but it seemed like the 545 could do the job. Still extremely pleased with the Leo Bodnar Mini GPSDO.

[CaptDon]

Don't make the assumption that if your scope is set at 1ms per division times
10 divisions that the horizontal sweep will be able to accurately trigger in perfect
synchronization to a 100hz horizontal input frequency and faithfully track a 200hz
vertical input. There is a delay in the horizontal timebase, both in the variable holdoff
which usually has some minimum which isn't zero and the time needed for retrace
in analog scopes. Your scope may accurately show the first 10 divisions equalling
the timespace of one cycle of a 100hz wave, but it will miss the trigger on the next
edge since the edge came before the scope was ready for it and now you will be
in a 'slipped' timeframe where exactly 100hz referenced against 200hz will appear
to show drift and seem to indicate the signals are not sychronis. As far as lissajous
patterns to be stable all you need is a common denominator even if you have to
convert the fraction to some higher order to find a common denominator such as
when you have to change 6/5 to 6000/5000 to suit the equation and you end up
with a weird split like 11/5 when you reduce to simple terms, basically whole numbers.
11.1/5 won't fly!!

[bob91343]

Complicating all of this is the possibility that one or both waves are not sinusoidal.  This results in patterns that are not in the book but you have to use ingenuity to decipher tham.